Author/Authors :
Adams، نويسنده , , Paul، نويسنده ,
Abstract :
It is proposed that vertebrate brains, especially those of mammals, operate according to an algorithm subsumable as “synaptic Darwinism”. The key postulate is that genes and synapses follow the same rules, because they act as autocatalytic, hypercyclic, units of selection. Synapses replicate by quantally strengthening, and mutate by connecting new cells. Because synapses relate pre- and post-synaptic firing, they perform a translation operation. Furthermore the product of this operation, conjoint firing, favors replication (by Hebbʹs Rule). The result is that variants are selected and patterns of connection automatically adopt optimal configurations. These configurations are determined by scalar neuromodulatory “reward” signals applied globally to layers of neurons, which reduce spike frequency adaptation and enhance Hebbian replication. Global or local control of mutation rates provides further improvements in the Darwinian algorithm. All the processes and circuits postulated have plausible, and often obvious, implementations. The result is that brains evolve and adapt like large ecosystems.